Re: tuptoaster.c must *not* use SnapshotAny
От | Hiroshi Inoue |
---|---|
Тема | Re: tuptoaster.c must *not* use SnapshotAny |
Дата | |
Msg-id | 3C47D33A.3A514CC8@tpf.co.jp обсуждение исходный текст |
Ответ на | Re: tuptoaster.c must *not* use SnapshotAny (Bruce Momjian <pgman@candle.pha.pa.us>) |
Список | pgsql-hackers |
Joe Conway wrote: > > Hiroshi Inoue wrote: > > > Though I've often seen the reference to bytea BLOB > > I remember no clear negation. Don't we have to negate > > it clearly from the first ? > > > > regards, > > Hiroshi Inoue > > With the changes in 7.2, bytea is actually very close to matching the > SQL99 definition of binary strings, which it also refers to as BLOB. > BLOBS should include support for (section 4.3): > > <comparison predicate> - yes > <blob concatenation> - yes > <blob substring function> - yes > <blob overlay function> - no > <trim function> - yes > <length expression> - yes > <position expression> - yes > <like predicate> - yes > Other than the overlay function, It seems good as long as they are not large(huge ?). It seems a significant flaw for bytea to have no possibility of partial update. > the primary thing missing is the use of > hex as the I/O representation. I don't love the current I/O representation of bytea. It seems worth adding a new data type only to change the binary I/O representation. regards, Hiroshi Inoue
В списке pgsql-hackers по дате отправления: